Learning to Model Multimodal Semantic Alignment for Story Visualization

TL;DR

This paper proposes a novel GAN-based approach for story visualization that dynamically aligns semantic levels between text and images, improving realism and consistency without auxiliary tools.

Contribution

It introduces a dynamic semantic alignment mechanism in GANs to better match text and image representations for story visualization.

Findings

Enhanced image quality and story consistency

Outperforms state-of-the-art methods

Effective without segmentation masks or auxiliary networks

Abstract

Story visualization aims to generate a sequence of images to narrate each sentence in a multi-sentence story, where the images should be realistic and keep global consistency across dynamic scenes and characters. Current works face the problem of semantic misalignment because of their fixed architecture and diversity of input modalities. To address this problem, we explore the semantic alignment between text and image representations by learning to match their semantic levels in the GAN-based generative model. More specifically, we introduce dynamic interactions according to learning to dynamically explore various semantic depths and fuse the different-modal information at a matched semantic level, which thus relieves the text-image semantic misalignment problem. Extensive experiments on different datasets demonstrate the improvements of our approach, neither using segmentation masks nor auxiliary captioning networks, on image quality and story consistency, compared with state-of-the-art methods.